[llvm] [RISCV] Improve contant materialization to end with 'not' if the cons… (PR #66950)

[Philip Reames via llvm-commits]

https://github.com/preames approved this pull request.

LGTM - trying the negate seems like a reasonable idea.  

As a follow up, could we use a sra to sign extend a leading one bit?  Would that ever be better than inverting?  

https://github.com/llvm/llvm-project/pull/66950